The purchasing power parity puzzle, temporal aggregation, and half-life estimation

We show, analytically, that ordinary least square (OLS)-based estimates of the half-life in a temporally aggregated AR(1) model suffer from large asymptotic biases. We then demonstrate, via simulations, that accurate estimates of half-lives can be obtained in finite samples using temporally aggregated data. (C) 2004 Elsevier B.V. All rights reserved.